Childhood amnesia — our inability to remember almost everything that happened to us when very young — is always interesting. It’s not as simple as an inability to form long-term memories. Most adults can’t remember events earlier than 3-4 years (there is both individual and cultural variability), even though 2-year-olds are perfectly capable of remembering past events (side-note: memory durability increases from about a day to a year from age six months to two years). Additionally, research has shown that young children (6-8) can recall events that happened 4-6 years previously.
Given that the ability to form durable memories is in place, what governs which memories are retained? The earliest memories adults retain tend to be of events that have aroused emotions. Nothing surprising about that. More interesting is research suggesting that children can only describe memories of events using words they knew when the experience occurred — the study of young children (27, 33 or 39 months) found that, when asked about the experimental situation (involving a "magic shrinking machine") six months later, the children easily remembered how to operate the device, but were only able to describe the machine in words they knew when they first learned how to operate it.
Put another way this isn’t so surprising: our memories depend on how we encode them at the time. So two things may well be in play in early childhood amnesia: limited encoding abilities (influenced but not restricted to language) may mean the memories made are poor in quality (whatever that might mean); the development of encoding abilities means that later attempts to retrieve the memory may be far from matching the original memory. Or as one researcher put it, the format is different.
A new study about childhood amnesia looks at a different question: does the boundary move? 140 children (aged 4-13) were asked to describe their three earliest memories, and then asked again two years later (not all could provide as many as three early memories; the likelihood improved with age).
While more than a third of the 10- to 13-year-olds described the same memory as their very earliest on both occasions, children between 4 and 7 at the first interview showed very little overlap between the memories (only 2 of the 27 4-5 year-olds, and 3 of the 23 6-7 year-olds). There was a clear difference between the overlap seen in this youngest group (4-7) and the oldest (10-13), with the in-between group (8-9) being placed squarely between the two (20.7% compared to 10% and 36%).
Moreover, children under 8 at the first interview mostly had no overlap between any of the memories they provided at the two interviews, while those who were at least 8 years old did. For the oldest groups (10-13), more than half of all the memories they provided were the same.
The children were also given recall cues for memories they hadn’t spontaneously recalled. That is, they were told synopses of memories belonging to both their own earlier memories, and other children’s earlier memories. Almost all of the false memories were correctly rejected (the exceptions mostly occurred with the youngest group, those initially aged 4-5). However, the youngest children didn’t recognize over a third of their own memories, while almost all the oldest children’s memories were recognized (90% by 8-11 year-olds; all but one by 12-13 year-olds). Their age at the time of the event didn’t seem to affect the oldest or the very youngest groups, but 6-9 year-olds were more likely to recall after cuing events that happened at least a year later than those events that weren’t recalled after cuing.
In general, the earliest memories were several months later at the follow-up than they had been previously. The average age at the time of the earliest memory was 32 months, and 39.6 months on the follow-up interview. This shift in time occurred across all ages. Moreover, for the very earliest memory, the time-shift was even greater: a whole year.
In connection with the earlier study I mentioned, regarding the importance of language and encoding, it is worth noting that by and large, when the same memories were recalled, the same amount of information was recalled.
There was no difference between the genders.
The findings don’t rule out theories of the role of language. It seems clear to me that more than one thing is going on in childhood amnesia. These findings bear on another aspect: the forgetting curve.
It has been suggested that forgetting in children reflects a different function than forgetting in adults. Forgetting in adults matches a power function, reflecting the fact that forgetting slows over time (as is often quoted, most forgetting occurs in the first 24 hours; the longer you remember something, the more likely you are to remember it forever). However, there is some evidence that forgetting in children is best modeled in an exponential function, reflecting the continued vulnerability of memories. It seems they are not being consolidated in the way adults’ memories are. This may be because children don’t yet have the cognitive structures in place that allow them to embed new memories in a dense network.
 . Infantile Amnesia Across the Years: A 2‐Year Follow‐up of Children’s Earliest Memories. Child Development [Internet]. 2011 ;82(4):1092 - 1105. Available from: http://onlinelibrary.wiley.com/doi/10.1111/j.1467-8624.2011.01597.x/abstract
 . Breaking the Barrier? Children Fail to Translate Their Preverbal Memories into Language. Psychological Science [Internet]. 2002 ;13(3):225 - 231. Available from: http://www.jstor.org/stable/40063711